CN103077391B - Car target localization method and device - Google Patents
Car target localization method and device Download PDFInfo
- Publication number
- CN103077391B CN103077391B CN201210591549.3A CN201210591549A CN103077391B CN 103077391 B CN103077391 B CN 103077391B CN 201210591549 A CN201210591549 A CN 201210591549A CN 103077391 B CN103077391 B CN 103077391B
- Authority
- CN
- China
- Prior art keywords
- car
- information area
- mentioned
- described information
- license plate
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of car target localization method and device.In the above-mentioned methods, License Plate is performed for input picture, obtain License Plate information; The above-mentioned License Plate information obtained is adopted to determine car target information area to be positioned; Adopt the characteristic parameter corresponding to circle to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, in above-mentioned information area, determining above-mentioned car target position to be positioned.According to technical scheme provided by the invention, the circular car target that can realize above to car plate is accurately located, and completes the effective identification to vehicle vehicle.
Description
Technical field
The present invention relates to computer image processing field, particularly a kind of car target localization method and device.
Background technology
Increasing along with socioeconomic development and vehicle, becomes a kind of inexorable trend by computer information, intelligently management vehicle.
License plate recognition technology is widely used in traffic flow monitoring, and highway bayonet socket is charged, in red light violation vehicle monitoring and community automatic fare collection system.Current treatment technology can only identify, but can not identify concrete vehicle car plate and large-scale, medium-sized, dilly.
Car mark shape common at present is mainly divided into circle, ellipse and rectangle.But in correlation technique, circular car target is accurately located and also lacks relevant technical scheme, therefore, for the circular car mark of car plate upper area, need a kind of effective technical scheme to locate accurately.
Summary of the invention
The present invention proposes a kind of car target localization method and device, cannot carry out pinpoint technical matters at least to solve in correlation technique to the circular car mark above car plate.
According to an aspect of the present invention, a kind of car target localization method is provided.
Car target localization method according to the present invention comprises: perform License Plate for input picture, obtains License Plate information; The above-mentioned License Plate information obtained is adopted to determine car target information area to be positioned; Adopt the characteristic parameter corresponding to circle to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, in above-mentioned information area, determining above-mentioned car target position to be positioned.
According to an aspect of the present invention, a kind of car target locating device is provided.
Car target locating device according to the present invention comprises: acquisition module, for performing License Plate for input picture, obtains License Plate information; Determination module, determines car target information area to be positioned for adopting the above-mentioned License Plate information of acquisition; Detection and location module, for adopting the characteristic parameter corresponding to circle to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, determining above-mentioned car target position to be positioned in above-mentioned information area.
Pass through the present invention, first determine car target general information region according to License Plate information phase, then in the fine positioning stage, the characteristic parameter corresponding to circle is adopted to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, in above-mentioned information area, determining above-mentioned car target position to be positioned.Solve in correlation technique and cannot carry out pinpoint technical matters to the circular car mark above car plate.Thus the circular car target that can realize above to car plate is accurately located, and completes the effective identification to vehicle vehicle.
Accompanying drawing explanation
Accompanying drawing described herein is used to provide a further understanding of the present invention, and form a application's part, schematic description and description of the present invention, for explaining the present invention, does not form inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is the process flow diagram of the car target localization method according to the embodiment of the present invention;
Fig. 2 is the process flow diagram of car target localization method according to the preferred embodiment of the invention;
Fig. 3 is the structured flowchart of the car target locating device according to the embodiment of the present invention; And
Fig. 4 is the structured flowchart of car target locating device according to the preferred embodiment of the invention.
Embodiment
Fig. 1 is the process flow diagram of the car target localization method according to the embodiment of the present invention.As shown in Figure 1, comprise the following steps (step S102-step S106) according to the car target localization method of the embodiment of the present invention:
Step S102: perform License Plate for input picture, obtains License Plate information;
Step S104: adopt the above-mentioned License Plate information obtained to determine car target information area to be positioned;
Step S106: adopt the characteristic parameter corresponding to circle to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, determine above-mentioned car target position to be positioned in above-mentioned information area.
In FIG, car target location is mainly divided into three phases.First stage (being equivalent to step S102): first car plate is positioned; Subordinate phase (being equivalent to step S104): determine car target general information region according to the License Plate information that the first stage obtains, subordinate phase (being equivalent to step S104): the i.e. accurate positioning stage of car target, the characteristic parameter corresponding to circle is adopted to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, in above-mentioned information area, determining above-mentioned car target position to be positioned.By effective combination of above-mentioned three steps, solve in correlation technique and cannot carry out pinpoint technical matters to the circular car mark above car plate.Thus the circular car target that can realize above to car plate is accurately located, and completes the effective identification to vehicle vehicle.
Wherein, in step s 102, be utilize the feature of license plate area to judge licence plate to the starting point of License Plate, license plate area is split from view picture vehicle image.Car plate self has many inherent features, according to the different characteristic of car plate, can adopt different localization methods.The method of current License Plate is a lot, modal vehicle license location technique mainly contain the method based on rim detection, the method based on Color Segmentation, the method based on wavelet transformation, the method based on genetic algorithm, based on the License Plate of mathematical morphology and the method etc. based on gray level image analysis of texture.
Wherein, for the method based on rim detection, so-called " edge " just refers to that its surrounding pixel gray scale has the set of those pixels of Spline smoothing.The both sides at " edge " belong to two regions, and the uniform gray level in each region is consistent, and the gray scale in these two regions characteristically exists certain difference.The task of rim detection accurately locates edge and restraint speckle.The method detected has multiple, such as Roberts boundary operator, Prewitt operator, Sobel operator and Lapalace edge detection.These methods utilize grey scale change this feature violent in object edge place to come the edge of detected image just.Each operator is different to the sensitivity of different edge type, and the effect of generation is also different, analyzes known through great many of experiments, and Roberts boundary operator is a kind of operator utilizing local variance operator to find edge, locates more accurate; Prewitt operator and Sobel operator have certain rejection ability to noise, but can not get rid of pseudo-edge completely; Laplace operator is Second Order Differential Operator, has rotational invariance to the step change type marginal point accurate positioning in image, but easily loses the directional information at a part of edge, and anti-noise ability is poor simultaneously.For different environment and requirement, suitable operator is selected to carry out to image the effect that rim detection just can reach.Be more than the simple description carried out the license plate locating method based on rim detection, other vehicle license location technique see the description in correlation technique, can repeat no more herein.
Wherein, in step S104, owing to locating car plate, therefore can position the car target general information region be positioned at above car plate according to the positional information of car plate.
In preferred implementation process, after execution step S104, before performing step S106, following process can also be comprised: to above-mentioned car target information area executive level correction process.
Intelligent transportation system (IntelligentTrafficSystem, referred to as ITS)
In, the licence plate of the object normally moving vehicle of picked-up, video camera generally can only be erected at above the side of highway, thus the license plate image collected has inclination and distortion in various degree, and the inclination of more than 3 degree can cause the obvious sex change of character, major part optical character identification (OpticalCharacterRecognition, referred to as OCR) method is difficult to adapt to, and this brings very large difficulty to the accurate segmentation of character and identification.Therefore, for this problem, can to above-mentioned car target information area executive level correction process.
In preferred implementation process, the process of above-mentioned execution above-mentioned rectification may further include following process:
(1) adopt maximum between-cluster variance OTSU algorithm by license plate binary, obtain binary map;
(2) in above-mentioned binary map, extract horizontal edge information, obtain horizontal edge figure;
(3) to above-mentioned horizontal edge figure in predetermined angular range at a predetermined angle interval carry out angle rotation, for each rotation, add up horizontal projection value maximum in four lines before above-mentioned horizontal edge figure, choosing maximum angle in this angle corresponding to horizontal projection value is car plate RA;
(4) bilinear interpolation algorithm is adopted to be corrected according to above-mentioned car plate RA by above-mentioned information area.
By above-mentioned process, in conjunction with the actual conditions of license plate image, harmless rectification is rotated to the horizontal direction of car plate, then adopt sciagraphy to carry out stretcher strain rectification in vertical direction, can effectively realize correcting the horizontal tilt of car target information area.
In preferred implementation process, above-mentioned steps S106 may further include following process:
(1) for above-mentioned information area, edge image is obtained by edge detection algorithm;
(2) on the above-mentioned edge image obtained, the characteristic parameter corresponding to above-mentioned circle (such as, circular feature amount etc.) is adopted to obtain maximum point in above-mentioned information area;
Wherein, circular feature C=μ
r/ δ
r, when region R trends towards circle, circular feature amount C monotone increasing and trend towards infinite, it is by the impact of region translation, rotation and dimensional variation, and C is a characteristic quantity with all frontier points definition of region R, wherein μ
rfor the mean distance from regional center to frontier point
(x in above-mentioned formula
k, y
k) be the coordinate of any point in image; (x, y) is regional barycenter coordinate; Definition δ
rfor the distance mean square deviation from regional barycenter to frontier point
(3) carry out loop truss by broad sense hough transform, determine that described car to be positioned is marked on the position of described information area;
(4) adopt mathematical morphology filter algorithm, locate above-mentioned car target coordinate to be positioned and in above-mentioned information area, intercept above-mentioned car mark to be positioned.
In preferred implementation process, after step s 104, before step S106, following process can also be comprised: gray processing process is performed to above-mentioned information area, and adopt structural element to perform opening operation to the gray level image obtained.
Wherein, opening operation belongs to morphological images process, is first corrode rear expansion, and its effect is: can make edge smoothing, eliminate tiny spine, disconnect narrow connection, keeps size constant.Use same structural element first to corrode the computing of expanding again to image and be called opening operation.Particularly, the opening operation under structural element B is defined as follows:
As can be seen here, opening operation can be used for eliminating small object thing, while the border of very thin some place separating objects, level and smooth larger object and its area of not obvious change.Therefore, after step s 104 before step S106, adopt opening operation, can effectively some irrelevant background images in car target information area be removed.
In preferred implementation process, after employing structural element performs opening operation to the gray level image obtained, adopt the characteristic parameter corresponding to above-mentioned circle to before above-mentioned information area execution loop truss, following process can also be comprised: adopt OTSU algorithm by above-mentioned Binary Sketch of Grey Scale Image, and obtain connected domain by morphological operation.
Wherein, so-called " binaryzation ", is set to 0 or 255 by the gray-scale value of the pixel on image exactly, namely whole image is presented obvious black and white effect.And obtain connected domain by morphological operation, there is a variety of method in the related, specifically see the description in correlation technique, can repeat no more herein.
Below in conjunction with Fig. 2, above-mentioned preferred implementation is further described.
Fig. 2 is the process flow diagram of car target localization method according to the preferred embodiment of the invention.As shown in Figure 2, this car target localization method can comprise following process:
Step S202: utilize rim detection or machine learning scheduling algorithm to carry out License Plate to input picture.
In preferred implementation process, location can be performed to car plate in the following ways:
Y
1=Ymax-t*Hpai
Y
2=Ymin-t*Hpai
X
1=Xmin
X
2=Xmax
Wherein, t=0.5 ~ 4, Y
1and Y
2car mark region up-and-down boundary coordinate respectively, X
1and X
2be the coordinate of region right boundary respectively, Ymax and Ymin is the coordinate of license plate area up-and-down boundary respectively, Xmin and Xmax is the coordinate of license plate area right boundary respectively, and Hpai is the height of car plate.
Step S204: according to License Plate information coarse positioning car target information area.
The region pre-service of step S206(car mark coarse positioning): rectification is carried out in the region (i.e. above-mentioned car target information area) of the parameter utilizing car plate to correct to car mark coarse positioning.
Particularly, the rotation angle information in car plate rectification can be utilized to carry out horizontal tilt correction in conjunction with bilinear interpolation to car mark region.
(1) extracting the horizontal edge hum pattern of binaryzation car plate, namely by row search binary map, when running into saltus step, when namely running into 0 → 1 or 1 → 0, will outline map be 1 with binary map white pixel corresponding point assignment.
(2) by the horizontal edge figure that extracts with 1
0for step-length is [-20
0, 20
0] horizontally rotate in scope, often rotate once, add up often row projection value in the horizontal direction respectively, get wherein maximum front n capable, then sue for peace, as the horizontal projection value (n is threshold value, can get 4) of this anglec of rotation.
(3) the horizontal projection value Sum (α) that tries to achieve of more each anglec of rotation, finds the maximum corresponding angle [alpha] of projection value to be the angle of car plate horizontal tilt.
(4) bilinear interpolation is utilized to carry out horizontal tilt correction to coarse positioning region with α angle.
Step S208: gray processing is carried out in the car mark coarse positioning region of correcting, utilizes structural element to carry out an opening operation to gray level image, for removing some irrelevant background images in car target information area.
Step S210: utilize OTSU Binarization methods to carry out binaryzation, then carry out morphological operation, obtain connected domain, removes white point very little for area in car target information area.
Step S212: first carry out circle detection by the how round fast algorithm of detecting of Generalized Hough Transform, detect round approximate location, then less circular constraint factor is set (such as, the constraint function of initial segmentation gets 0 usually) adopt the numerical solution of partial differential equation after suitable iteration, the approximate contours of all targets in image can be extracted, as the initial segmentation of target, now adopt level set function segmentation image, make in each image, to comprise a target object.
For each target object, the distance reinitialized with above-mentioned segmentation level set function is for initial setting up, larger circular constraint function is set (such as, can 200 be got) to detect circular object, then adopt partial differential equation to solve it, if the solution of trying to achieve is 0, then target object is circular object, then all target images are stacked up, just obtain all circular targets in original image.
If the circle of being detected as, then perform step S214, otherwise, can arrange and perform one or many detection again, in case deviation appears in testing result, improve and detect correctness.Fig. 2 shows the technical scheme performing one-time detection again, if it is not circular for namely detecting, performs step S218.
Step S214: detect connected region according to axis, will the circular connected region label for labelling of axis be positioned at, and process respectively, count the position of connected domain, thus realize car target and accurately locate.
Step S216: at coarse positioning and correct after car target information area in intercept final car mark.
Step S218: whether secondary detection car mark is circular; If so, step S214 is performed.Otherwise flow process terminates.
Fig. 3 is the structured flowchart of the car target locating device according to the embodiment of the present invention.As shown in Figure 3, this car target locating device includes but not limited to lower module:
Acquisition module 30, for performing License Plate for input picture, obtains License Plate information;
Determination module 32, is connected with acquisition module 30, determines car target information area to be positioned for adopting the above-mentioned License Plate information of acquisition;
Detection and location module 34, be connected with determination module 32, for adopting the characteristic parameter corresponding to circle to perform loop truss to above-mentioned information area, determining that above-mentioned car to be positioned is designated as circular car timestamp, in above-mentioned information area, determining above-mentioned car target position to be positioned.
By effective combination of three modules of the vehicle-logo location device shown in Fig. 3, solve in correlation technique and cannot carry out pinpoint technical matters to the circular car mark above car plate.Thus the circular car target that can realize above to car plate is accurately located, and completes the effective identification to vehicle vehicle.
In preferred implementation process, as shown in Figure 4, said apparatus can also comprise: rectification module 36, is connected between determination module 32 and detection and location module 34, for above-mentioned information area executive level correction process.
In preferred implementation process, as shown in Figure 4, above-mentioned rectification module 36 may further include: the first acquiring unit 360, for adopting maximum between-cluster variance OTSU algorithm by license plate binary, obtains binary map; Second acquisition unit 362, for extracting horizontal edge information in above-mentioned binary map, obtains horizontal edge figure; Statistic unit 364, for above-mentioned horizontal edge figure in predetermined angular range at a predetermined angle interval carry out angle rotation, for each rotation, add up horizontal projection value maximum in four lines before above-mentioned horizontal edge figure; Choosing unit 366, is car plate RA for choosing angle maximum in the angle corresponding to this horizontal projection value; Correcting unit 368, corrects above-mentioned information area according to above-mentioned car plate RA for adopting bilinear interpolation algorithm.
In preferred implementation process, as shown in Figure 4, above-mentioned detection and location module 34 may further include: the 3rd acquiring unit 340, for for above-mentioned information area, obtains edge image by edge detection algorithm; 4th acquiring unit 342, for the above-mentioned edge image obtained, adopts the characteristic parameter corresponding to above-mentioned circle to obtain maximum point in described information area; Determining unit 344, for carrying out loop truss by broad sense hough transform, determines that above-mentioned car to be positioned is marked on the position of described information area; Positioning unit 346, for adopting mathematical morphology filter algorithm, locating above-mentioned car target coordinate to be positioned and in above-mentioned information area, intercepting above-mentioned car mark to be positioned.
In preferred implementation process, said apparatus also comprises: the first processing module 38, can being connected between rectification module 36 and detection and location module 34, being connected to for performing gray processing process to above-mentioned information area, and adopt structural element to perform opening operation to the gray level image obtained.
In preferred implementation process, said apparatus also comprises: the second processing module 40, can be connected between the first processing module 38 and detection and location module 34, for adopting OTSU algorithm by above-mentioned Binary Sketch of Grey Scale Image, and obtains connected domain by morphological operation.
It should be noted that, each module in above-mentioned car target locating device, and the preferred working method be combined with each other of each unit, see the description in Fig. 1 to Fig. 2, can repeat no more herein.
To sum up above-mentioned, by above-described embodiment provided by the invention, the circular car target that can realize above to car plate is accurately located, and completes the effective identification to vehicle vehicle.Further, by adopting some optimization process schemes, such as, to car target information area rectification, the extraneous background image etc. that opening operation removes car target information area is performed, can more accurately positioning car mark, identify vehicle.
Through the above description of the embodiments, those skilled in the art can be well understood to the mode that the present invention can add required general hardware platform by software and realize, and can certainly pass through hardware, but in a lot of situation, the former is better embodiment.Based on such understanding, technical scheme of the present invention can embody with the form of software product the part that prior art contributes in essence in other words, this computer software product can be stored in storage medium, as ROM/RAM, magnetic disc, CD etc., comprising some instructions in order to make a computer equipment (can be personal computer, server, or the network equipment etc.) perform the above-mentioned method of some part of each embodiment of the present invention or embodiment.
Below be only the preferred embodiment of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.
Claims (8)
1. a car target localization method, is characterized in that, comprising:
License Plate is performed for input picture, obtains License Plate information;
The described License Plate information obtained is adopted to determine car target information area to be positioned;
Adopt the characteristic parameter corresponding to circle to perform loop truss to described information area, determining that described car to be positioned is designated as circular car timestamp, in described information area, determining described car target position to be positioned;
Wherein, adopt the characteristic parameter corresponding to circle to perform loop truss to described information area, determining that described car to be positioned is designated as circular car timestamp, in described information area, determining that described car target position to be positioned comprises:
For described information area, obtain edge image by edge detection algorithm;
On the described edge image obtained, the characteristic parameter corresponding to described circle is adopted to obtain maximum point in described information area;
Carry out loop truss by broad sense hough transform, determine that described car to be positioned is marked on the position of described information area;
Adopt mathematical morphology filter algorithm, locate described car target coordinate to be positioned and in described information area, intercept described car mark to be positioned;
Wherein, circular feature parameter C meets formula: C=μ
r/ δ
r;
Wherein μ
rfor the mean distance from regional center to frontier point; δ
rfor the distance mean square deviation from regional barycenter to frontier point;
(x
k, y
k) be the coordinate of any point in image; (x, y) is regional barycenter coordinate.
2. method according to claim 1, is characterized in that, before the characteristic parameter corresponding to employing circle performs loop truss to described information area, also comprises: to described information area executive level correction process.
3. method according to claim 2, is characterized in that, performs described rectification process and comprises:
Adopt maximum between-cluster variance OTSU algorithm by license plate binary, obtain binary map;
Described binary map is extracted horizontal edge information, obtains horizontal edge figure;
To described horizontal edge figure in predetermined angular range at a predetermined angle interval carry out angle rotation, for each rotation, add up horizontal projection value maximum in four lines before described horizontal edge figure, choosing maximum angle in this angle corresponding to horizontal projection value is car plate RA;
Bilinear interpolation algorithm is adopted to be corrected according to described car plate RA by described information area.
4. according to the method in any one of claims 1 to 3, it is characterized in that, before the characteristic parameter corresponding to the described circle of employing performs loop truss to described information area, also comprise:
Gray processing process is performed to described information area, and adopts structural element to perform opening operation to the gray level image obtained;
Adopt OTSU algorithm by described Binary Sketch of Grey Scale Image, and obtain connected domain by morphological operation.
5. a car target locating device, is characterized in that, comprising:
Acquisition module, for performing License Plate for input picture, obtains License Plate information;
Determination module, determines car target information area to be positioned for adopting the described License Plate information of acquisition;
Detection and location module, for adopting the characteristic parameter corresponding to circle to perform loop truss to described information area, determining that described car to be positioned is designated as circular car timestamp, determining described car target position to be positioned in described information area;
Described detection and location module comprises:
3rd acquiring unit, for for described information area, obtains edge image by edge detection algorithm;
4th acquiring unit, for the described edge image obtained, adopts the characteristic parameter corresponding to described circle to obtain maximum point in described information area;
Determining unit, for carrying out loop truss by broad sense hough transform, determines that described car to be positioned is marked on the position of described information area;
Positioning unit, for adopting mathematical morphology filter algorithm, locating described car target coordinate to be positioned and in described information area, intercepting described car mark to be positioned;
Wherein, circular feature parameter C meets formula: C=μ
r/ δ
r;
Wherein μ
rfor the mean distance from regional center to frontier point; δ
rfor the distance mean square deviation from regional barycenter to frontier point;
(x
k, y
k) be the coordinate of any point in image; (x, y) is regional barycenter coordinate.
6. device according to claim 5, is characterized in that, also comprises: rectification module, for described information area executive level correction process.
7. device according to claim 6, is characterized in that, described rectification module comprises:
First acquiring unit, for adopting maximum between-cluster variance OTSU algorithm by license plate binary, obtains binary map;
Second acquisition unit, for extracting horizontal edge information in described binary map, obtains horizontal edge figure;
Statistic unit, for described horizontal edge figure in predetermined angular range at a predetermined angle interval carry out angle rotation, for each rotation, add up horizontal projection value maximum in four lines before described horizontal edge figure;
Choosing unit, is car plate RA for choosing angle maximum in the angle corresponding to this horizontal projection value;
Correcting unit, corrects described information area according to described car plate RA for adopting bilinear interpolation algorithm.
8. the device according to any one of claim 5 to 7, is characterized in that, also comprises:
First processing module, for performing gray processing process to described information area, and adopts structural element to perform opening operation to the gray level image obtained;
Second processing module, for adopting OTSU algorithm by described Binary Sketch of Grey Scale Image, and obtains connected domain by morphological operation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210591549.3A CN103077391B (en) | 2012-12-30 | 2012-12-30 | Car target localization method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210591549.3A CN103077391B (en) | 2012-12-30 | 2012-12-30 | Car target localization method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103077391A CN103077391A (en) | 2013-05-01 |
| CN103077391B true CN103077391B (en) | 2016-01-20 |
Family
ID=48153915
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210591549.3A Expired - Fee Related CN103077391B (en) | 2012-12-30 | 2012-12-30 | Car target localization method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103077391B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104391966B (en) * | 2014-12-03 | 2017-09-29 | 中国人民解放军国防科学技术大学 | Typical logo searching method based on deep learning |
| CN105160300B (en) * | 2015-08-05 | 2018-08-21 | 山东科技大学 | A kind of text abstracting method based on level-set segmentation |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1928892A (en) * | 2006-09-20 | 2007-03-14 | 王枚 | Method and device for license plate location recognition, vehicle-logo location recognition and vehicle type |
| CN101196980A (en) * | 2006-12-25 | 2008-06-11 | 四川川大智胜软件股份有限公司 | Method for accurately recognizing high speed mobile vehicle mark based on video |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050027541A1 (en) * | 2003-07-31 | 2005-02-03 | Glen Hagood | System and Method for Fabricating Informational Placard |
-
2012
- 2012-12-30 CN CN201210591549.3A patent/CN103077391B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1928892A (en) * | 2006-09-20 | 2007-03-14 | 王枚 | Method and device for license plate location recognition, vehicle-logo location recognition and vehicle type |
| CN101196980A (en) * | 2006-12-25 | 2008-06-11 | 四川川大智胜软件股份有限公司 | Method for accurately recognizing high speed mobile vehicle mark based on video |
Non-Patent Citations (3)
| Title |
|---|
| 基于PCA和边缘不变矩的车标识别新方法;王玫等;《计算机工程与应用》;20080229;第224页第2节、图1-图5 * |
| 质量退化的车牌字符分割方法;李文举等;《计算机辅助设计与图形学学报》;20040531;第698页第2.2节，图3-图4 * |
| 车标定位方法研究;李玲;《中国优秀硕士论文全文数据库 信息科技辑》;20100715;第I138-839第29页第3.6节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103077391A (en) | 2013-05-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1796043B1 (en) | Object detection | |
| US8848978B2 (en) | Fast obstacle detection | |
| US7792327B2 (en) | Apparatus and method for detecting a road boundary | |
| JP5787024B2 (en) | Three-dimensional object detection device | |
| CN107169494B (en) | License plate image based on handheld terminal divides bearing calibration | |
| Jimenez et al. | Traffic sign shape classification and localization based on the normalized FFT of the signature of blobs and 2D homographies | |
| Yan et al. | A method of lane edge detection based on Canny algorithm | |
| CN103942797B (en) | Scene image text detection method and system based on histogram and super-pixels | |
| CN109784344A (en) | A kind of non-targeted filtering method of image for ground level mark identification | |
| CN105488501A (en) | Method for correcting license plate slant based on rotating projection | |
| Wang et al. | A vision-based road edge detection algorithm | |
| Ying et al. | Robust lane marking detection using boundary-based inverse perspective mapping | |
| CN104794421A (en) | QR (quick response) code positioning and recognizing methods | |
| US10235576B2 (en) | Analysis method of lane stripe images, image analysis device, and non-transitory computer readable medium thereof | |
| EP1796042B1 (en) | Detection apparatus and method | |
| CN103065140A (en) | Location method and device for automobile logos | |
| CN102999886A (en) | Image edge detector and ruler raster grid line precision detection system | |
| CN105303153A (en) | Vehicle license plate identification method and apparatus | |
| CN103065144A (en) | Automobile logo identification method and device | |
| JP5011268B2 (en) | Image detection apparatus and image detection method | |
| Li et al. | Road markings extraction based on threshold segmentation | |
| CN108256445A (en) | Method for detecting lane lines and system | |
| CN106327464A (en) | Edge detection method | |
| CN103077391B (en) | Car target localization method and device | |
| CN104732530A (en) | Image edge detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160120 Termination date: 20191230 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |